New York Times Sues Brave Over 30-Billion-Page Index Sold to AI Labs

The New York Times and three Tribune Publishing newspapers sued Brave Software in Manhattan federal court on November 2, alleging that the browser and search company copied their journalism into a web index of more than 30 billion pages and sold access to it to AI developers. The complaints name Anthropic, Perplexity, Mistral, xAI and Cohere as customers of the Brave Search API, though none of the AI companies is a defendant.

Brave's search API is the target of two copyright suits in Manhattan

Brave crawls the web to build its own search index, uses it to power Brave Search and its Ask Brave AI answers, and sells access to the same index through a paid API that AI companies use to ground chatbot answers in live web results and, the publishers allege, to train models. The Tribune papers bring five claims: direct, contributory and vicarious copyright infringement, circumvention of technical protections under the Digital Millennium Copyright Act, and unjust enrichment under New York law. The Times brings its own copyright and DMCA claims and says it sent Brave cease-and-desist letters in August 2025 and April 2026.

The publishers lean on Brave's own words. When Brave launched its Leo assistant on November 2, 2023, the company wrote that Anthropic used Brave's Search API to train its Claude 2 model, a sentence both complaints cite. The suits also target Brave's paywall-blocking setting and its default blocking of publisher ads.

“The Brave Search crawler does not advertise a differentiated user agent because we must avoid discrimination from websites that allow only Google to crawl them.”

Brave Search, crawler help page

That design choice sits at the center of the case. Brave says its bot will skip any page that Googlebot is barred from, but a publisher that wants to stay in Google while keeping out an AI data reseller has no separate name to block. The publishers argue this leaves them no practical way to opt out of Brave's index.

Publishers are now suing the data layer under AI search

Earlier publisher cases went after the model makers directly: the Times has sued OpenAI and Microsoft, and Perplexity. These suits aim at the vendor in between, and they arrive two days after OpenAI, Google, Microsoft and Anthropic backed a shared robots.txt token for opting out of AI training. A token only works for crawlers that announce themselves, which is why Brave's unnamed crawler is the publishers' sharpest allegation.

Brave had not responded when the suits were filed, but it has argued in a separate case that its indexing is fair use. It sued News Corp in March 2025 seeking that ruling, and News Corp countersued for infringement in July 2026. The first US appellate ruling on AI training, Thomson Reuters v. Ross, went against the AI defendant, and Brave now faces two courts weighing whether selling an index for model training is search or resale.

For AI labs the exposure is indirect: discovery could show which models were trained or grounded on Brave data, and an injunction would cut off one of the few independent web indexes outside Google and Bing that they can license today.